[yt-users] Problem Running yt in Parallel

Matthew Turk matthewturk at gmail.com
Fri Jul 29 08:32:19 PDT 2011


Hi Greg,

Thanks for writing.  This looks familiar to me, and I think it is
fixed in the current development branch.  Can you use the dev module
for yt instead of stable?

One sidenote is that often slicing in parallel is not a net gain; I
have found very few datasets where it is in fact substantially faster,
most of them gigantic AMR runs with >1e5 grids.  I would recommend
just doing slices in serial.

Best,

Matt

On Fri, Jul 29, 2011 at 11:15 AM,  <meecegre at msu.edu> wrote:
> Hi everyone,
>
> I've come across an error while trying to use yt for parallel analysis on
> Kraken. Wondering if its something people have seen before.
>
> I have my simple script that does slices of a few different quanties. When I
> ran it with 12 tasks on Kraken, it worked fine for the first four datasets,
> but crashed during the fifth one. I tried a simpler script that just does a
> slice of density for that dataset, and it crashed again with the same error:
>
> IndexError: arrays used as indices must be of integer (or boolean) type
>
> When I ran this code again with only 1 task (still on Kraken, everything
> else the same), it worked fine. More of the output, as well as my job script
> and plotting script, are below.
>
> Since it works in serial , and works for some data sets, the problem seems
> to be something to do with parallelism. Is there any sort of simple
> explanation or something I'm doing wrong?
>
> Just wondering if this is something people have come across before.
>
> --Greg
>
> (Last few lines of output)
> mg, mc, mv, pos = self.find_max_cell_location(field, finest_levels)
> mg, mc, mv, pos = self.find_max_cell_location(field, finest_levels)
> File
> "/lustre/scratch/proj/sw/yt/current/lib/python2.7/site-packages/yt-2.1stable-py2.7-linux-x86_64.egg/yt/data_objects/object_finding_mixin.py",
> line 69, in find_max_cell_location
> File
> "/lustre/scratch/proj/sw/yt/current/lib/python2.7/site-packages/yt-2.1stable-py2.7-linux-x86_64.egg/yt/data_objects/object_finding_mixin.py",
> line 69, in find_max_cell_location
> File
> "/lustre/scratch/proj/sw/yt/current/lib/python2.7/site-packages/yt-2.1stable-py2.7-linux-x86_64.egg/yt/data_objects/object_finding_mixin.py",
> line 69, in find_max_cell_location
> max_grid = self.grids[mg]
> IndexError: arrays used as indices must be of integer (or boolean) type
> max_grid = self.grids[mg]
> max_grid = self.grids[mg]
> IndexError: arrays used as indices must be of integer (or boolean) type
> IndexError: arrays used as indices must be of integer (or boolean) type
>
> (Job script)
> #!/bin/bash
> #PBS -A TG-AST090040
> #PBS -l size=12,walltime=2:00:00
> #PBS -j oe
> #PBS -N slicer
>
> cd /lustre/scratch/oshea/nested_ics_real1/analysis
>
> ls
> pwd
>
> export MPLCONFIGDIR=${PBS_O_WORKDIR}/.matplotlib/
> [ ! -d ${MPLCONFIGDIR} ] && mkdir ${MPLCONFIGDIR}
>
> module unload PrgEnv-pgi
> module load PrgEnv-gnu
> module load yt
>
> aprun -n 12 python ./simple_slice.py --parallel
>
>
> (simple_slice.py)
> from yt.mods import *
>
> DD = 40
>
> fileName = "../DD%s/DD%s" % ((str(DD)).zfill(4),  (str(DD)).zfill(4))
>
> pf = load(fileName)
> pc = PlotCollection(pf)
>
> pc.add_slice("Density", 0)
>
> pc.save("./%s" % str(DD).zfill(4))
>
>
>
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>
>



More information about the yt-users mailing list